CHAPTER 1
Defects and Microstructures in Feldspars
BY A. C. McLAREN
1 Introduction
The minerals of the feldspar group are probably the most important of all rock-forming substances since they make up between 50 and 60 weight per cent of all igneous rocks and, in addition, they occur under a wide range of geological conditions. In fact, the classification of rocks is based to a large extent on the quantity and kinds of feldspars present.
In view of this, the feldspars have been studied in greater detail than any other group of minerals. The literature is, therefore, voluminous but fortunately there are good summaries in many elementary text books on mineralogy, as well as more extended reviews. In addition, there is the monumental work by J. V. Smith.
Optical microscopy and X-ray diffraction studies in particular have shown that the feldspars are an extremely complicated group of minerals and that specimens as we find them are very rarely homogeneous single crystals with grown-in or stress-induced crystals defects such as dislocations and twins. In general, feldspar specimens consist of a complex microstructure which is the product of order-disorder and structural transformations, as well as diffusion controlled processes such as exsolution (solid state precipitation). In addition, feldspar specimens which have been strained (deformed) in response to externally applied stresses develop characteristic micro-structures.
Before proceeding further it is necessary to consider briefly the nature of the feldspar series of minerals and their basic crystal structure.
The feldspars fall into two main series: (i) the alkali feldspars KA1Si3O8 to NaA1Si308 and (ii) the plagioclase feldspars NaA1Si3O8 to CaA12 Si2O8. These end members are referred to as orthoclase (Or), albite (Ab), and anorthite (An), respectively.
The chemical composition of any feldspar mineral is usually given in terms of the mol per cent of Or, Ab, and An. The alkali feldspars usually contain less than IO per cent An, but the Na-rich members (such as anorthoclase) may contain more. Similarly, the plagioclase feldspars usually contain less than IO per cent Or. The composition of any specimen is written as AnxAbyOrz: where x,y, and z are the concentrations in mol per cent.
The feldspar structure is based on a framework of (Al,Si)O4 tetrahedra, with the metal ions (K, Na, Ca) occupying positions in the interstices of the framework. The idealized feldspar structure is monoclinic with space group C2/m and there is complete disorder in the occupancy of tetrahedral sites by Al and Si. This is the structure of the K-rich mineral sanidine which has 4 KA1Si3O8 per unit cell. However, the fully ordered structure of KA1Si3O8 is triclinic, with space group C]bar.1]. This mineral is known as (maximum) microcline and has a structure similar to that of low-temperature fully ordered ablite.
With the exception of monalbite (a high-temperature, monoclinic, disordered form of albite), the plagioclase feldspars are all triclinic. Because the Al/Si ratio in anorthite (An100) is 1:1, ordering requires a regular alternation of Al and Si in the framework and this produces a body-centred structure I[bar.1] with a doubled c-axis. For this reason, it is usual when considering the plagioclase series to use this larger unit cell containing eight formular units.
The structural type exhibited by any particular feldspar specimen and its unit cell dimensions are clearly a function of the chemical composition and the degree of ordering which itself is dependent upon the temperature of crystallization and the subsequent thermal history of the specimen. For example, feldspars in volcanic rocks which have crystallized at high temperature followed by quenching to a low temperature may retain their high-temperature disordered state. On the other hand, feldspars in rocks which have cooled slowly may become ordered into a low-temperature state.
However, when any feldspar specimen is cooled from a high temperature, other significant changes may also occur, as indicated above. It is these changes, together with changes directly related to ordering, which are responsible for the micro-structures observed. Sometimes the microstructure is on a coarse enough scale for it to be observed directly by optical microscopy. For example, alkali feldspar specimens of intermediate composition which are homogeneous at high temperatures, exsolve at low temperatures into periodically-twinned domains of Na-rich feldspar and untwinned-domains of K-rich feldspar. On the other hand, the existence in some specimens of microstructures of exsolution and/or twinning on a very fine scale was implied originally from X-ray observations. But such observations provide little or no information about the size, shape, and distribution of the domains.
The first successful use of transmission electron microscopy to obtain such information on an alkali feldspar was made by Fleet and Ribbe. They showed that a K-rich moonstone from Ceylon consisted of alternating lamellae of monoclinic orthoclase and triclinic albite approximately parallel to (601). The lamellae were of the order of 1000 Å wide and the albite lamellae were periodically twinned on the albite-twin law. These observations gave a detailed explanation of the features of the associated diffraction pattern and further explained the origin of the well-known 'schiller' (or interference colours) exhibited by moonstone.
These observations were, of necessity, made on the thin edges of tiny crushed fracture fragments of the moonstone. In spite of the obvious disadvantages of this method of specimen preparation, many useful TEM studies of feldspars and other minerals were made. However, most of these disadvantages have been overcome by the development of ion-bombardment thinning. With this technique it became possible to produce consistently extensive thin areas of a wide range of non-metallic materials, and to relate TEM and optical microscope observations directly. As a consequence, there has been a spectacular increase in the application of TEM to mineralogical and petrological problems. McLaren has reviewed the TEM observations of feldspars which had been carried out up to the beginning of 1972. Since then a number of important observations have been made. In the plagioclase feldspars extensive work has been carried out on (i) antiphase domains; (ii) the coexistence of domains of different structural type; (iii) exsolution; iv) the occurrence and nature of the superlattice structure in specimens of intermediate composition (An25 to An75); and (v) crystal defects and their role in the mechanisms of deformation.
Work on the alkali feldspars has been almost exclusively concerned with exsolution phenomena, in particular spinodal decomposition and its growth into incoherent precipitates, together with the role of twinning in reducing stress. this work was done on alkali feldspars in the composition range from about Or 30Ab70 to about Or70Ab30, and many details of the nature of the observed microstructures and their origin are now well understood.
However, in spite of an extensive background of optical microscope and X-ray studies, only two TEM studies have been made recently on specimens outside this composition range where microstructures associated with the monoclinic to triclinic transition are expected to be conspicuous. Both these studies were concerned with microcline (Or8 [+ or -] 10) and they must be considered, essentially, as preliminary.
It would, I believe for two reasons, be inappropriate and, indeed, unnecessary to attempt a comprehensive review of all these observations here. Firstly, the true significance of much of the TEM work cannot be appreciated without an understanding of the mineralogical background which needs to be developed in some detail for each particular problem. Secondly, excellent and critical reviews covering much work done over the last five years have recently been given by Heuer and Nord, Champness and Lorimer, Yund, Ribbe, and, of course, J. V. Smith. The problem here has been to select those topics for discussion which will produce a review whose character is (i) sufficiently synoptic to interest a wide range of materials scientists (chemists, physicists, mineralogists, and geologists) in the application of TEM to solid state problems and (ii) sufficiently seminal to interest people actively engaged in feldspar research.
In an attempt to satisfy these criteria I have decided to restrict this review to a detailed discussion of two problems of current interest, each showing clear evidence for the necessity of further work. The first of these problems is concerned with the nature of the superlattice structure responsible for the e- and f-reflections in the plagioclase feldspars of intermediate composition (e-plagioclases). The most recent work' on this problem has involved the use, for the first time in feldspar research, of high-resolution lattice-imaging techniques. The second problem is concerned with the defects and microstructures associated with the monoclinic to triclinic transition in the alkali feldspars. This work is particularly concerned with transformation twinning and exsolution in microcline (Or > 80 per cent) and anorthoclase (Or < 30 per cent), and illustrates the combined use of conventional optical and electron microscope techniques.
2 The Structure of e-Plagioclase
High-resolution lattice-imaging TEM has recently been applied to the long-standing problem of the nature of the e-plagioclases. In this section an attempt will be made to review these observations critically in the light of earlier work using TEM and X-ray diffraction. However, this must be preceded by some background mineralogical and crystallographic considerations, as well as a discussion of the microstructures of specimens with the e-plagioclase structure.
Background. — The plagioclase feldspars exhibit three basic structural types: (i) the albite structure C]bar.1] with a-reflections (h+k = 2n;l = 2n) only, (ii) the body-centred structure I]bar.1] with a-reflections and b-reflections (h+ k = 2n + 1; l=2n + 1), and (iii) the primitive structure P]bar.1] with c-reflections (h + k = 2n ; 1 = 2n+1) and d-reflections (h + k = 2 n+1; I = 2 n) in addition to a- and b-reflections. Also, there is the superlattice (or e-plagioclase) structure which is characterized by pairs of e-reflections in place of b-reflections, and f-reflections as satellites to a-reflections. There are no c- or d-reflections. A typical diffraction pattern is shown in Figure 1. McLaren has analysed this diffraction pattern and shown that both the e- and f-reflections are due to a superlattice with period T= 1/|t|; where t is the reciprocal lattice vector giving the position of an f-satellite relative to the associated a-reflection. Vector t also defines the position of one e-reflection of a pair relative to the other e-reflection.
This superlattice structure is observed only in specimens in the intermediate composition range (An25 to An75) which have been cooled slowly from high temperatures. Both the magnitude and direction of t vary with An-content. These variations were first determined from X-ray diffraction patterns, but electron diffraction has recently been used to obtain more accurate measurements over the range An34 to An69 in a cogenetic series of specimens. When the direction of t is plotted on a stereographic projection, the observations for various An-contents lie, in general, on a great circle. Similarly, the observed values of the magnitude of t as a function of An-content lie approximately on a straight line. However, there is evidence to suggest that t does not vary continuously across the intermediate plagioclase range, there being a discontinuity at a composition of about An50. Also, the rate of change with An-content of both the magnitude and direction of t increases with higher An-content. Thus, any error in the An-content will introduce a greater uncertainty in t for specimens of higher An-content. Both the direction and magnitude of t observed by McLaren in An 75 have been questioned, but they are smooth extrapolations of the curves quoted above, T= l/|t| being 85 Å. In a specimen of estimated composition An72, Heuer and Nord (personal communication) found T= 96 Å, which is certainly much higher than expected. On the other hand, Grove quotes values of T as low as 37 Å in specimens of An76. It is clear that at compositions around An75 there is considerable uncertainty about the magnitude and direction of t.
The problem with e-plagioclases is a two-fold one: Firstly it is necessary to determine the basic structure, and secondly to understand the observed variation of t with An-content. However, before going on to consider the structure, it is necessary to investigate the microstructure of specimens which exhibit the characteristic diffraction pattern.
Microstructure. — The two specimens showing the e-plagioclase diffraction pattern which have been examined by high-resolution lattice-imaging TEM are both schiller labradorites with compositions of An54 and An52. The microstructure of similar specimens has also been examined in detail by TEM, and these observations will now be discussed.
Although there is no evidence in normal X-ray or electron diffraction patterns for exsolution in these specimens, TEM has revealed the existence of a microstructure consisting of two kinds of alternating lamellae of thicknesses of the order of 1000 Å. It is this microstructure which is responsible for the interference colours or Schiller exhibited by these specimens. Subsequent observations using the analytical electron microscope have shown that the lamellae differ in composition by about 12 per cent An.
The lamellae may be visible in bright-field (BF) images but they are better seen and more easily analysed in dark-field (DF) images. The diffraction contrast of the lamellae in DF images with a-reflections has been studied in detail by McLaren and it was shown that the contrast arises because of a slight misorientation between adjacent lamellae. When the crystal is in the symmetrical position on the rocking curve (i.e. when the mean deviation from the exact Bragg angle s= 0) the contrast is extremely weak and there is a reversal of contrast with the sign of s, as occurs across a bend contour. It was also found that one set of lamellae (A) is consistently wider than the other set (B). The observed contrast indicates that the intensities of a-reflections in A and B [and hence the structure factors Fa(A) and Fa(B)] are not significantly different at s= 0.
However, this is not so for e-reflections. It can be seen from the DF image of Figure 2 that the diffraction contrast is extremely strong at s = 0. The A and B lamellae are bright and dark, respectively. There is no reversal of contrast with the sign of s. This indicates that the structure factors for e-reflections, Fe(A) and Fe(B), are significantly different, in fact it appears that Fe(B) = 0. It follows that the 30 Å superlattice associated with the e- and f-reflections should be observed in A-lamellae only. This has been confirmed by the direct resolution of the superlattice (i) using a pair of e-reftections as in Figure 3 and (ii) using the 020,000,020 and the six associated f-reflections as in Figure 10 of McLaren. Since no (unsplit) b-reflections are observed in the diffraction patterns, it must be concluded that the B-lamellae have the high-temperature (disordered) structure C]bar.1].
